Regulation of localization and activity of the microtubule depolymerase MCAK

Abstract

Mitotic Centromere Associated Kinesin (MCAK) is a potent microtubule depolymerizing and catastrophe-inducing factor, which uses the energy of ATP hydrolysis to destabilize microtubule ends. MCAK is localized to inner centromeres, kinetochores and spindle poles of mitotic cells, and is also present in the cytoplasm. Both in interphase and in mitosis, MCAK can specifically accumulate at the growing microtubule ends. Here we discuss the mechanisms, which modulate subcellular localization and activity of MCAK through the interaction with the End Binding (EB) proteins and phosphorylation.

Introduction

Microtubules (MTs) are dynamic polymers, which alternate between phases of growth and shortening, a behavior known as dynamic instability.¹ The spontaneous transitions between polymerization and depolymerization can occur in solutions of purified tubulin and are powered by the energy of tubulin-driven GTP hydrolysis. In cells, each type of MT transition can be controlled by specialized regulatory factors. Among such factors, a prominent position is occupied by MT-destabilizing enzymes of the kinesin-13 family.^2,3 While most kinesins use their motor domain to translocate along MTs, kinesin-13 family members employ their ATP-hydrolyzing motor domain to destabilize MT ends. Unlike other kinesins, which have their motor at the N- or C-terminus of the molecule, kinesin-13s have a centrally located motor domain preceded by the neck region (Fig. 1). A kinesin-13 fragment encompassing the neck and the motor is sufficient for MT-depolymerizing activity,^4,5 while the N- and the C-terminal domains are important for protein dimerization and subcellular localization.

Figure 1.

A scheme of MCAK protein sequence. Protein domains and phosphorylation sites are indicated.

Four kinesin-13 proteins are encoded in the human genome, KIF2A, KIF2B, KIF2C (MCAK) and KIF24, of which MCAK is the best-studied family member. MCAK (also known as XKCM1 in Xenopus) participates in the assembly and dynamics of mitotic spindle MTs and is thought to contribute to correct chromosome segregation through control of kinetochore-MT attachments.^6–11 Loss of MT-destabilizing activity of MCAK leads to excessive elongation of astral MTs,^7,12 defects in spindle bipolarity^13–15 and an increased incidence of lagging chromosomes during anaphase, possibly due to incorrect kinetochore-MT attachments induced by the decreased turnover of kinetochore-MTs.^9,16 Also in interphase cells, MCAK can reduce MT stability and make MTs more dynamic.¹⁵

Overexpression of MCAK leads to complete depolymerization of MTs.¹⁷ In physiological conditions, the access of MCAK to MT ends is likely to be controlled, and a key question is how the localization and the activity of MCAK are regulated depending on the phase of the cell cycle. Studies based on in vitro reconstitution and cell biological approaches are beginning to shed light on the regulatory mechanisms controlling MCAK-dependent MT destabilization.

Targeting of MCAK to the Growing MT Tips

MCAK can potently depolymerize MTs that are stabilized by taxol and/or slowly hydrolysable GTP analogue GMPCPP, and this is the MCAK activity that has been most extensively studied in vitro using purified MCAK protein.^4,18–20 These studies showed that MCAK has a high affinity for MT ends, that it can target MT tips by lateral diffusion along MT lattice²⁰ and that the MT on-rate, dependent on the positively charged neck region, is a key parameter regulating MCAK depolymerase activity.¹⁹

Importantly, in cells MCAK usually acts in conditions when MT ends are dynamic and decorated by numerous additional regulators. In physiological contexts, MT depolymerization can proceed spontaneously, and therefore the function of MCAK is not to assist disassembly of shrinking MTs but to trigger catastrophes of growing or stabilized MT ends. In cells, all polymerizing MT plus ends are bound by a variety of proteins known as MT plus end tracking proteins or +TIPs.^21,22 These proteins form comet-like accumulations at the distal 1–2 µm of the growing MT. At the core of the +TIP complex are End-Binding (EB) proteins, which autonomously recognize growing MT ends.²³ Recently, it was shown that this property is likely to be due to the preference of EB proteins for the freshly assembled GTP-bound tubulin (GTP cap) at the MT plus end.²⁴

EBs can recruit to growing MT ends numerous additional factors, most of which are MT stabilizers.²³ Paradoxically, MCAK is also an EB-interacting +TIP, and this property is conserved in invertebrates, underscoring its functional significance.^25,26 The relationship between EBs and MCAK is complex: in cells, EBs can suppress MT catastrophes independently of their interactions with their +TIP partners,²⁷ and an excess of EB1 antagonizes MT depolymerizing activity of MCAK.²⁶ At the same time, EBs are required for MCAK association with growing MT plus ends.²⁸ Two possibilities could be envisioned for the function of MCAK-EB interaction at the growing MT tips. First, this interaction could keep MCAK inactive at MT end until its activity is somehow triggered.²⁸ In this scenario, an MCAK mutant that would be unable to bind to EBs would be more efficient in destabilizing MTs compared to EB-binding MCAK present at the same concentration at the MT tip. Alternatively, concentration of MCAK near the plus end could enhance the chance of a productive interaction with MT tips and promote depolymerization of EB-bound MT plus ends. In this case, depolymerizing activity of MCAK would depend primarily on its abundance at the tip, irrespectively of its capacity to directly interact with EBs, while the EB binding would simply serve as a mechanism to concentrate MCAK at the MT plus ends.

Recent advances made it possible to directly address these possibilities. First, the MT plus end tracking phenomenon has been reconstituted in vitro using purified proteins, including mammalian EBs, enabling the analysis of the effects of purified +TIPs on MT dynamics.^29,30 Second, the detailed structural basis of the interactions between EB proteins and some of their partners, including MCAK, has been elucidated,²³ making it possible to specifically disrupt these interactions and thus question the function of MT tip tracking behavior without destroying other functional properties of a given +TIP.

MCAK represents an ideal candidate to address the function of EB-dependent MT tip-tracking because it has a very potent activity (MT depolymerization) that can be easily measured both in vitro and in cells. MCAK binds to the EB proteins through a so-called SxIP motif—a positively charged polypeptide region rich in serines and prolines containing the sequence Ser-any amino acid-Ile-Pro (SxIP, which reads SKIP in MCAK) that is located in the MCAK N-terminus³¹ (Fig. 1). Similar motifs are present in several +TIPs, including the spectraplakin MACF2 and the tumor suppressor APC, and the structure of the complex of the SxIP-containing peptides with the C-terminal cargo binding domain of EB1 has been determined by X-ray crystallography and NMR.³¹ This structural study showed that the isoleucine and proline residues within the SxIP motif are important for the interaction with a hydrophobic cavity within the C-terminal region of EB1 (Fig. 2), while the positive charges within the peptides contribute to the electrostatic interaction with the negatively charged surface of the EB1 C-terminal domain. The interaction is also stabilized by a network of hydrogen bonds formed between the EB1 C-terminus and the SxIP peptide (Fig. 2). The importance of the SxIP motif in MCAK is underscored by the fact that its two close homologues, KIF2A and KIF2B, have no such sequence and cannot bind to EB1 or track polymerizing MT plus ends.

Figure 2.

A scheme of the interaction of SxIP-motif containing peptides with EB1. CH, calponin homology domain; CC, coiled coil and four-helix bundle (EB1 dimerization domain).

Based on the knowledge of the structure of SxIP-EB complex, three strategies were devised to disrupt it.^31,32 First, the Ile-Pro binding hydrophobic cavity is in part shaped by the otherwise unstructured C-terminal tail of EB1 (Fig. 2). Removal of this tail disrupts the cavity and effectively abolishes SxIP binding without affecting the N-terminal calponin homology domain of EB1, which is responsible for MT end recognition.^24,27,33 Second, the SxIP-EB1 interaction depends on formation of hydrogen bonds, which involve EB1 residues Tyr217 and Glu225 (Fig. 2). Mutation of these amino acids to alanines (YE/AA double mutation) strongly destabilizes the SxIP-EB1 complex. Finally, since Ile and Pro residues of MCAK are essential for the interaction with the hydrophobic cavity in EB1 C-terminus (Fig. 2), substitution of these MCAK residues for hydrophilic asparagines (the IP/NN double mutation) abolishes the interaction with EB1 without affecting the MT-depolymerizing domain of MCAK. Application of these strategies made it possible to investigate whether an EB protein is necessary and sufficient to recruit MCAK to MT tips and what is the function of such recruitment.

Purified GFP-tagged MCAK was combined with purified mCherry-tagged EB3 in a MT plus end tracking in vitro assay.³² EB3 was chosen for this experiment instead of EB1 because, unlike EB1, it tolerates an N-terminal fluorescent tag, which allows generation of a fluorescent fusion that would not interfere with the cargo binding at the C-terminus. Reconstitution experiments showed that, in line with cellular studies, MCAK was less potent in depolymerizing EB-decorated MTs as compared to “naked” MTs. Further, EB3 was able to recruit MCAK to growing MT tips, but only when it could directly bind to MCAK. In these conditions, MCAK increased MT catastrophe frequency without affecting MT depolymerization rate, again consistent with its cellular function. However, in all cases when MCAK could not bind to EB3, i.e., when either EB3 without the tail or the EB3 YE/AA mutant was combined with the wild type MCAK, or when the MCAK IP/NN mutant was used together with wild type EB3—much more MCAK was needed to induce the same level of MT destabilization than when wild type MCAK and wild type EB3 were used. Thus, in all cases, the destabilizing activity of MCAK depended on its accumulation at the MT tip, and this accumulation was much higher when MCAK could directly interact with EB3. Also in cells, in the presence of endogenous EB proteins, the wild type MCAK was a much more potent depolymerase than the MCAK mutant that is unable to bind to EB proteins.³²

Taken together, these results indicate that the interaction with EBs enhances the activity of MCAK by helping MCAK to concentrate near the MT end. In the crowded cellular environment, a direct interaction with EBs also likely helps MCAK to gain access to MT tips and to overcome the activities of MT stabilizing and polymerization-promoting factors, such as XMAP215/ch-TOG.^34,35

Are the EBs the only regulators that directly affect MCAK interaction with the growing MT ends? MCAK binds to EBs in the same way as multiple other SxIP proteins,²³ and therefore it would likely undergo competition with these proteins. Competition for the limited space at the MT tip is a general property of the +TIP interaction network.²¹ +TIPs deal with it by undergoing additional, often redundant interactions with each other. Remarkable in this respect is the interaction of MCAK with TIP150 (known in Xenopus as ICIS, a protein that stimulates MT-depolymrizing activity of MCAK at inner centromeres).^36,37 TIP150 is also an SxIP containing +TIP, which can enhance the interaction of MCAK with MT ends in an EB-dependent manner.³⁷ Furthermore, in Xenopus, MCAK was reported to interact with APC,³⁸ which also binds to EB1 through SxIP motifs.^31,39 These interactions are likely to regulate the activity of MCAK at the MT tips in certain cellular contexts.

Another important question concerns the kinetics of MCAK at the MT tips. In mammals, EB proteins and their partners do not stay associated with the growing MT ends: while tubulin subunits are added to the MT plus end, +TIPs, including MCAK, undergo rapid cycles of binding and unbinding within the 1–2 µm stretch of the freshly polymerized MT lattice.^29,30,32,40 The interaction with EB3 increased the association rate of MCAK with EB-decorated MT tips, confirming the importance of this kinetic parameter for MCAK activity.^19,32 The dwell time of MCAK molecules at MT tips was also somewhat increased by the interaction with EB3.³² Interestingly, the dwell time is longer for MCAK than for EB3 itself, suggesting that once MCAK is recruited to the MT end, it can remain associated with it even if its EB3 partner is released. Single molecule imaging showed that in conditions when MCAK tracks growing MT ends with the aid of EB3, lateral diffusion of MCAK along the MT lattice is relatively low.³² Still, it is possible that also under these conditions the diffusion along the lattice contributes to MCAK delivery to the outmost MT tip, where MCAK would induce a catastrophe. It is also possible that in cells additional factors promote MCAK delivery to the outmost MT tips.

Regulation of MCAK by Phosphorylation

The complex manner in which MCAK interacts with MT ends provides numerous potential ways to regulate it. MCAK is a target for multiple kinases (Fig. 1 and Table 1). Among them, the best studied is Aurora B, which can phosphorylate several sites within the N-terminus and the neck region of MCAK. The addition of phosphates decreases the positive charge of these regions and is expected to negatively affect the interaction of MCAK with MTs and EB1. Indeed, phosphorylation of serines in the vicinity of the SxIP motif in the N-terminus of MCAK disrupts EB binding and plus end tracking activity of MCAK.^26,31 In addition, the interaction of MCAK with TIP150 is reduced by Aurora B-dependent phosphorylation,³⁷ and this might contribute to diminished recruitment of MCAK to growing MT ends and thus attenuate MT-depolymerizing activity of MCAK.

Table 1.

Identified phosphorylation sites in MCAK from mammals and Xenopus

Residue	Human residue	Domain	Kinase	Effect of phosphorylation on in vitro activity	Localization of phosphorylation	In vivo function of phosphorylation	Organism: MCAK/experimental system
S92 (Cg)6	S95	N-term	Aurora B		Cytoplasm; Kinetochores-asymmetrically distributed over kinetochore pair6 and higher on kinetochores with merotelic attachments.11	Phosphorylation of S95, S109, S111 inhibits MCAK binding to EB1.26,31 MCAK-S95E, S192E double mutant has reduced depolymerization activity in vivo. MCAK-S95E, S109E, S111E, S115E, S192E (5E) has further reduced activity. MCAK-5E localizes to inner centromeres, while MCAK-5A localizes to kinetochores.6 MCAK-5E also has reduced binding to TIP150.37	Hamster/Human cells
T95 (Xl)42				No effect	G2 phase-Chromosomes Prometaphase-Chromosomes and centromeres/kinetochores	MCAK-S95A has reduced chromosome arm binding and increased centromeres binding.42 MCAK-4A (S95, S110 (Xl), S166, S192) can reconstitute for wild type MCAK in MT depolymerization in vivo, but not in bipolar spindle assembly.41	Xenopus/Xenopus
S106 6	S109	N-term	Aurora B	n.d.		Phosphorylation of S95, S109, S111 inhibits MCAK binding to EB1.26,31 MCAK-S95E, S192E double mutant has reduced depolymerization activity in vivo. MCAK-S95E, S109E, S111E, S115E, S192E (5E) has further reduced activity. MCAK-5E localizes to inner centromeres, while MCAK-5A localizes to kinetochores.6 MCAK-5E also has reduced binding to TIP150.37	Hamster/Human cells
108 6	S111	N-term	Aurora B	n.d.		Phosphorylation of S95, S109, S111 inhibits MCAK binding to EB1.26,31 MCAK-S95E, S192E double mutant has reduced depolymerization activity in vivo. MCAK-S95E, S109E, S111E, S115E, S192E (5E) has further reduced activity. MCAK-5E localizes to inner centromeres, while MCAK-5A localizes to kinetochores.6 MCAK-5E also has reduced binding to TIP150.37	Hamster/Human cells
S110 41–42	N.c.	N-term	Aurora B	n.d.		MCAK-S110A has strongly reduced centromere binding.42 MCAK-S110 cannot fully replace wild type MCAK in chromosome alignment.42 MCAK-4A (S95, S110 (Xl), S166, S192) can reconstitute for wild type MCAK in MT depolymerization in vivo, but not in bipolar spindle assembly.41	Xenopus/Xenopus
S112 6	S115	N-term	Aurora B	n.d.		MCAK-S95E, S192E double mutant has reduced depolymerization activity in vivo. MCAK-S95E, S109E, S111E, S115E, S192E (5E) has further reduced activity. MCAK-5E localizes to inner centromeres, while MCAK-5A localizes to kinetochores.6 MCAK-5E also has reduced binding to TIP150.37	Hamster/Human cells
S161 10,41	N.c.	N-term	Aurora B	n.d.			Xenopus/Xenopus
T162 10	S152	N-term	Aurora B	n.d.			Xenopus/Xenopus
S177 10,41	S166	N-term	Aurora B	n.d.		MCAK-4A (S95, S110 (Xl), S166, S192) can reconstitute for wild type MCAK in MT depolymerization in vivo, but not in bipolar spindle assembly.41	Xenopus/Xenopus
S19610,41–42 (S186 in Cg)6	S192	Neck	Aurora B	Inhibition	In prometaphase-centromeres and small amount at centrosomes. The phosphorylation is strongly reduced in metaphase. In anaphase/telophase, pS196 (Xl) is observed in the MT midzone.	MCAK-S196A has a stronger affinity for chromosome arms.42 MCAK-S196A can substitute wild type MCAK for spindle assembly, but not for chromosome alignment. MCAK-S196E can not substitute for wild type MCAK in spindle assembly due to decreased MT depolymerization activity.42 MCAK-S95E, S192E double mutant has reduced depolymerization activity in vivo. MCAK-S95E, S109E, S111E, S115E, S192E (5E) has further reduced activity. MCAK-5E localizes to inner centromeres, while MCAK-5A localizes to kinetochores.6 MCAK-5E also has reduced binding to Tip150.37 MCAK-4A (S95, S110 (Xl), S166, S192) can reconstitute for wild type MCAK in MT depolymerization in vivo, but not in bipolar spindle assembly.41	Xenopus/Xenopus
T229 10	S225	Neck	Aurora B	n.d.			Xenopus/Xenopus
S253 10	T249	Neck	Aurora B	n.d.			Xenopus/Xenopus
S196 43	S192	Neck	Aurora A	Inhibition			Xenopus/Xenopus
S719 43	S715	C-term	Aurora A	No effect		MCAK-S719E has reduced binding to aster poles, but increased binding to spindle poles.43 MCAK-S719E has increased activity in promoting spindle bipolarity.43	Xenopus/Xenopus
T537 44		Motor domain	CDK1	Decreased MT depolymerization, but similar ATPase activity.	Spindle and poles	MCAK-T537E has decreased MT depolymerization activity in vivo, but non-phosphorylatable mutants are not more active.44 MCAK-T537A localizes stronger to the centrosomes.44 Both MCAK-T537A and T537E mutants cannot properly rescue MCAK RNAi in chromosome alignment.44	Human
S592, S595, S621, S633, S715 45		C-term	Plk1	n.d.		MCAK-5E mutants have reduced MT depolymerization activity in vivo.45 MCAK-5E mutants show enhanced intramolecular interactions between N-terminal and C-terminal domains.45	Human
??			CaMKIIγ	No effect		CaMKIIγ appears to suppress MCAK activity in vivo, but the mechanism is unknown.46	Human

Sites shown in bold are the preferred phosphorylation sites of Aurora B kinase. All amino acid numbers in columns describing function and localization of phosphorylation sites refer to residues in human MCAK, unless stated otherwise. Xl, Xenopus laevis; Cg, Cricetulus griseus; n.d., not determined; n.c., not conserved.

Phosphorylation in the positively charged MCAK neck region, which can be induced by Aurora B, but also Aurora A, can potentially affect the electrostatic interaction of this domain with the negatively charged MT lattice and reduce the MCAK MT on-rate. Consistent with this, phosphorylation of MCAK by Aurora kinases in the neck region inhibits the ability of MCAK to depolymerize MTs in vitro.^6,10,41 Phosphorylation of the neck region of MCAK is also observed in vivo and appears to be tightly regulated in space and time; it is high at centromeres early in mitosis, but much lower when chromosomes have fully aligned. Furthermore, while MCAK localizes robustly to centrosomes, MT plus-ends, centromeres and kinetochores, phosphorylation of MCAK in the neck region predominantly occurs at centromeres and to a lesser extent at spindle poles.¹⁰

Several N-terminally located MCAK phosphorylation sites control the binding of MCAK to spindle poles, centromeres, kinetochores and chromosome arms (see Table 1).^6,10,41–43 While it is clear that Aurora A and B control targeting of MCAK to all these sites, the exact mechanism by which these kinases, especially Aurora B, regulate MCAK in space and time is highly complex and not fully understood. For example, Aurora B-dependent phosphorylation of MCAK at S95 (in Xenopus MCAK) strongly inhibits MCAK binding to centromeres, while phosphorylation of MCAK at S110 by the same kinase increases binding to centromeres.⁴² Therefore, it is likely that Aurora B phosphorylates different sites at specific times and locations, but the mechanism underlying such specificity still needs to be resolved.

An inhibitory phosphorylation of MCAK within its motor domain has also been reported.⁴⁴ This modification depends on CDK1/cyclin B and, similar to Aurora kinase phosphorylation, is restricted to mitosis. In vitro experiments revealed that motor phosphorylation of MCAK by CDK1/cyclin B inhibits the intrinsic MT depolymerase activity of MCAK (although MCAK ATPase activity is not inhibited). Consistent with this, a phospho-mimicking mutant of MCAK shows reduced MT depolymerizing activity in vivo.⁴⁴ It should be noted that, since MT stability needs to be decreased in mitosis compared to interphase, it is unclear why CDK1 would inhibit MCAK activity specifically during cell division. Surprisingly, the MCAK mutant that cannot be phosphorylated at the CDK1 site does not have an increased MT depolymerase activity in mitosis, suggesting that only a small fraction of the MCAK pool might be phosphorylated at this site during cell division. Perhaps CDK1, like Aurora kinases, inhibits only a specific pool of MCAK, rather than the entire population, thereby further contributing to the spatial regulation of MT dynamics. The exact role of CDK1-mediated regulation of MCAK requires further elucidation.

Another layer of regulatory complexity stems from the fact that the C-terminal tail of MCAK can regulate its MT depolymerase activity, possibly through autoinhibition.¹⁷ The C-terminus of MCAK can be phosphorylated by Plk1.⁴⁵ Interestingly, while Aurora and CDK1 negatively regulate MCAK's activity, Plk1 appears to promote the activity of MCAK by phosphorylating its C-terminus. Although it is still unclear exactly how Plk1-dependent phosphorylation of MCAK stimulates its MT destabilizing activity, it is noteworthy that MCAK engages in several interor intramolecular interactions, which are affected by Plk1 phosphorylation.⁴⁵ Additional work will be needed to resolve the function of these interactions and to gain insight into the spatiotemporal Plk1-dependent regulation of MCAK.

Conclusions and Future Directions

MCAK has emerged as a highly potent but tightly controlled MT depolymerizing enzyme. The accumulation of MCAK at the growing MT plus ends, the major sites of MT-destabilizing activity of this kinesin-13 family member, depends on the EB proteins, which are positive regulators of MT growth. In vitro reconstitution approaches combined with the structure-function analysis made it possible to dissect the significance of the plus end tracking behavior of MCAK and provided a basis for similar studies of other +TIPs. These studies showed that EB-dependent recruitment of MCAK to the growing MT ends stimulates its activity and explained why the disruption of this interaction by phosphorylation inhibits MCAK's MT destabilizing function in cells.

Phosphorylation affects the intrinsic MT depolymerizing activity of MCAK, as well as its association with multiple subcellular sites in mitotic cells. The list of kinases that are able to phosphorylate MCAK, as well as the list of amino acids within the MCAK sequence that are modified is steadily growing. This knowledge will form the basis for elucidation of molecular interactions responsible for MCAK recruitment to different structures and for understanding of the elaborate spatiotemporal control of these interactions and the MT-destabilizing activity of MCAK.